Doomed, doomed.     As advertised in my September 22nd podcast, I have been listening, via Audible, to Mustafa Suleyman’s The Coming Wave, just published this month. If you’ll excuse me repeating myself:

The subtitle is: ”Technology, Power, and the Twenty-first Century’s Greatest Dilemma.” It’s about the, yes, coming wave of technology—most particularly Artificial Intelligence and bio-tech—that will transform the world we live in.

Suleyman, that Turkish-sounding name notwithstanding, is another Brit (although he now lives in Silicon Valley), and one well qualified to pronounce on his subject. He was a co-founder of DeepMind back in 2010, and has been a leader in AI research ever since. He has interesting and important things to say …

He sure does. No one knows better than Suleyman how fast things are moving in those two areas of technological advance.

That phrase in the book’s subtitle, ”the Twenty-first Century’s Greatest Dilemma,” refers to the great good these changes could accomplish, the great evil they might unleash, and how we steer the human race towards the good and away from the evil. The latter part of that—steering us away from the evil—Suleyman discusses at much length under the heading ”containment.”

The central problem for humanity in the 21st century is how we can nurture political power and wisdom, technical mastery, and robust norms to constrain technology and ensure they continue to do far more good than harm …

But:

The odds are stacked against us in making this a reality.

For a dyed-in-the-wool pessimist like me, the book is a banquet. Has the COVID panic got you worrying about possible leaks from big, expensive labs staffed by credentialed scientists under careful surveillance? Very soon, perhaps already, a hobbyist tinkering with synthetic biology in his garage could kill a billion of us. (Suleyman is actually quoting a biotech expert there, but he seems to believe it.)

Suleyman did have me shaking my head a few times, to the degree my headset would allow it. I suspect he is uncritically Woke. In Chapter 14, for example, we get this:

A few years ago many Large Language Models had a problem. They were, to put it bluntly, racist. Users could quite easily find ways to make them regurgitate racist material, or hold racist opinions they had gleaned while scanning the vast corpus of text on which they had been trained. Toxic bias was, it had seemed, ingrained in human writing and then amplified by AI.

This led many to conclude that the whole setup was ethically broken, morally nonviable. There was no way LLMs could be controlled well enough to be released to the public, given the obvious harms.

But then LLMs, as we’ve heard, took off. In 2023 it’s now clear that, compared with the early systems, it’s extremely difficult to goad something like ChatGPT into racist comments.

Is it a solved problem? Absolutely not; there are still multiple examples of biased, even overtly racist, LLMs, as well as serious problems with everything from inaccurate information to gaslighting. But for those of us who have worked in the field from the beginning, the exponential progress at eliminating bad outputs has been incredible, undeniable.

Suleyman doesn’t give us his definition of ”racist,” but I suspect it is congruent with Ibram X. Kendi’s. Thence to the question: What happens to truth—truth that contradicts social dogma—when AI supplies us with all our knowledge?

I’ll guess that LLM knowledge bases strongly resemble Wikipedia: handy if you want to look up Dirichlet’s Theorem or the Battle of Lepanto but deeply unreliable on anything—or anybody—connected to social dogma on race or sex.

And Western World Wokery is a pale, weak thing by comparison with social dogmas elsewhere. What would a Chinese LLM have to say about Mao Tse-tung’s great famine, or the Tiananmen Square protests? What would a North Korean LLM tell me about the Kim family?

And with North Korea in mind: If one guy fiddling in his garage can kill a billion of us, what might a malevolent nation do?

Mustafa Suleyman does his best to offer hope, scolding what he calls ”techno-pessimists” and urging us to more enlightened regulation, better international cooperation, etc. He sounds worried, though. So he should. ”Nurture political power and wisdom”? Has Mustafa Suleyman ever attended a session of the U.S. Congress?

Twenty years ago I reviewed a book by Britain’s Astronomer Royal arguing that humanity will not survive to see the 22nd century. He made a good case. The Coming Wave, for all the author’s earnest, plainly heartfelt urgings towards ”containment,” makes an even better one.

[Permalink]

The love that never fails.     In my July Diary I had words to say about romantic love and the decline thereof. I omitted to mention the deepest, most robust, most enduring variety of love. Permit me to remedy the omission.

Actually, I’m going to summon up Alexander Pushkin to remedy it for me. Here we are in Chapter 4 of Eugene Onegin. Tatyana has confessed to Onegin that she is in love with him. Onegin has given her an honest but brutally frank rejection, breaking the poor girl’s heart. The poet addresses his reader.

But whom to love? To trust and treasure?
Who won’t betray us in the end?
And who’ll be kind enough to measure
Our words and deeds as we intend?
Who won’t sow slander all about us?
Who’ll coddle us and never doubt us?
To whom will all our faults be few?
Who’ll never bore us through and through?
You futile, searching phantom-breeder,
Why spend your efforts all in vain?
Just love yourself and ease the pain,
My most esteemed and honoured reader!
A worthy object! Never mind,
A truer love you’ll never find.\

Eugene Onegin, 4:22, Falen’s translation. Original Russian here.

[Permalink]

The value of forms.     Issues of Congressional decorum were in the news this month. Thirty-six-year-old Representative Lauren Boebert (R, Colorado), who is in the throes of a divorce, was imaged getting seriously handsy with a date at a pop-music show: ”Roman hands and Russian fingers,” as the old joke goes. (You have to hear it spoken.)

And then in mid-September Senate Majority Leader Chuck Schumer annulled the Senate’s dress code for members. The junior senator from Pennsylvania promptly entered the chamber September 20th in shorts, beach shoes, and a short-sleeved, untucked leisure shirt with no tie.

There have been consequences in both cases. Rep. Boebert is of course up for reelection a year from now. Voters in her district are reported to be widely unhappy about what I guess we should call Gropegate.

And in the Senate chamber, a resolution that business attire be worn on the floor of the chamber was introduced on the 27th of the month by Joe Manchin of West Virginia, whom God preserve! The resolution requires men to wear a coat, tie, slacks and other long pants in the chamber. It was adopted by unanimous consent.

Senate Passes ‘SHORTS‘ Resolution; John Fetterman Must Wear Pants https://t.co/OeOsLOlqcB via @BreitbartNews

— Joel Pollak (@joelpollak) September 28, 2023

As a reactionary geezer I naturally favor dress codes. I am at one with the bishop in Barchester Towers, of whom Trollope tells us:

He understood well the value of forms, and knew that the due observance of rank could not be maintained unless the exterior trappings belonging to it were held in proper esteem.

I can recall the time when dress codes were everywhere. As a young computer programmer circa 1970 I had the idea to go for a job interview at IBM, which had a regional headquarters in west London. Asking around for advice, I heard from more than one informant that if I did go for interview I should on no account wear brown shoes. IBM didn’t hire applicants wearing brown shoes.

Now that’s a dress code! I commend it to the attention of Senators Manchin and Schumer.

[Permalink]

One small step for man.     Not only are my days numbered, so are my steps.

When we moved to our present house we acquired a dog. I took to walking the dog every day. Thirty-one years and two dogs later, I’m still doing it.

I soon structured my dog-walking to four routes appropriate to how much time I had on any given day. The four were: very short, short, regular, long. Then, to get a feel for how much walking I was doing, I counted my paces. My pace measures eighty centimeters on level ground.

(How do I know that? Because the local high school is nearby—810 paces away, if you must know—and has very nice sports facilities, including a running track marked off in 400-meter lanes. One day I hiked over there and paced off five times round one of the tracks—two thousand meters. It was 2,500 paces, near enough, and the result follows.)

From that I worked out the mileage on my four dog-walking routes. With paces to three-digit accuracy:

• Very short, when I’m really pressed for time: 1,290 paces, which is 0.62 miles.
  
  
• Short, when I have time but not much energy: 1,810 paces, = 0.90 miles.
  
  
• Regular, most weekdays: 3,260 paces, = 1.62 miles.
  
  
• Long, most Saturdays: 4,340 paces, = 2.16 miles.

(Mrs. Derbyshire walks the mutt on Sundays.)

Those computations were all done back in the Clinton administrations. I counted the paces off by … counting, in my head. People tell me that nowadays there is a smartphone app that’ll do the counting for you.

I still don’t have a smartphone nor any intention of getting one; but I’m glad to know the damn things are good for something besides making phone calls … and making totalitarian population control way, way easier than it’s ever been before.

[Permalink]

Hubble and Schrödinger, a century on.     While technology roars ahead at, according to Mustafa Suleyman, an accelerating rate, key regions of theoretical physics can’t seem to make any headway.

For science geeks this is especially frustrating right now. This present decade, the 2020s, is of course a century on from the 1920s. It was in the 1920s physicists suddenly, massively, improved their understanding of the big and the small.

I mean the really big and the really small. In that marvelous decade a hundred years ago we developed new, durable models of not only the biggest thing that physics talks about, the entire universe, but also the smallest things: those subatomic particles of which human-scale matter and energy are mere epiphenomena.

Imagine that a precise hundred years ago, in September 1923, you had asked an astronomer to describe the universe. He would most likely have told you that our own Sun, with all the planets and stars you can see in the sky, and some luminous smears and blobs—probably just gas clouds—mostly visible only in telescopes, all cohabit in a single huge disk-shaped collection.

The Milky Way, he would have told you, is our view of it all from our position inside the disk. There is nothing else, just this almighty star-disk. The Milky Way is the universe.

A Midwestern-American astronomer named Edwin Hubble, working in California, blew that all apart in 1924. Yes, indeed: all the stars in the sky and millions of others are packed into a disk-like structure many thousand trillion miles across—a galaxy. And yes: the Milky Way is that galaxy, seen from the inside.

However, some of those luminous smears and blobs are not part of our galaxy. They are galaxies in their own right, ”island universes,” some much bigger than the Milky Way, all millions or billions of trillions of miles away.

That not only revolutionized our understanding of the cosmos at large, it also led quickly to the overall model we still have: the Big Bang 13.8 billion years ago, the cosmic background radiation from that event, and the expanding universe—the galaxies (mostly) flying away from each other.

At the other end of the size scale the 1920s gave us mature Quantum Theory. To quote myself, in reference to the years 1925-27:

These were the years when it dawned on researchers that the intuitions we acquire through our interactions with reality at everyday scales of measurement are simply not appropriate to events in the realm of electrons and protons.

Those new understandings gained in the 1920s had great staying power. They are still our basic models for the colossally large and the inconceivably small. There have of course been modifications and improvements, but the cosmology and the microphysics of today resemble those of 1930 far, far more closely than 1930's resemble 1830's, or even 1920's.

There are also problems, though. What, exactly, keeps the universe expanding? All that matter in all those galaxies: shouldn’t its gravitational pull slow down and eventually reverse the expansion, ending in a Big Crunch? So it was supposed until the 1990s, when observations showed that the expansion is accelerating. Eh? What could cause that?

A hypothetical Dark Energy, a kind of energy hitherto unknown, makes the mathematical equations work; but then, according to quantum theory, we should be able to find a particle for Dark Energy, like the photon for electromagnetic energy, but … we can’t.

What about gravity? How does that look from the level of quantum theory? Is there a particle for it, a graviton? If there is, we can’t find it; and there are good reasons to think there isn’t.

And as if we didn’t have enough unresolved questions in cosmology to ponder, this wonderful new James Webb Space Telescope, now deep into its second year of operation, is giving us more. We can now see galaxies far, far back in that 13.8-billion-year span (because their light has taken billions of years to reach us). They don’t look the way our theories say they should have looked in the early universe, that close to the Big Bang.

Quantum theory itself still has intractable conundrums, like the Measurement Problem—the one underlying the famous Schrödinger’s cat thought-experiment. Efforts here have generated all sorts of oddities, most notably the Many Worlds interpretation, which has itself begotten nontrivial problems in Philosophy of Science. Is a theory any good if there is no conceivable way we might validate it?

So while it hasn’t by any means been total stasis in physics since Hubble and Schrödinger, it’s been walking pace by comparison to the amazing sprints of a hundred years ago, with even a few backward steps. Some of the problems they raised have proven as durable as the truths they revealed.

Perhaps AI will be able to untangle it all for us.

[Permalink]

Unreasonable anger.     Speaking of extinctions: When I hear that someone I know has committed suicide, I naturally feel sad. The sadness, however, is garnished with a light seasoning of anger—anger towards the deceased.

My anger is unreasonable, of course. It’s also by no means universal. Most people, the handful of times I’ve mentioned it in company, have looked puzzled at me, then looked away.

I’m not totally alone in it, though. When I mentioned it once to a work colleague, he confessed he felt the same. Then he offered the only explanation I’ve heard that is at all plausible. He: ”It’s like you’re in a game of football [i.e., soccer; this was in England] when all of a sudden one of your teammates stops playing and just walks off the field.”

[Permalink]

Math Corner.     Here’s one from Southall & Pantaloni’s charming little 2018 book Geometry Snacks: Bite Size Problems & How to Solve Them.

Prove that the three dark matchstick heads are collinear (that is, they lie on a single straight line).

[Permalink]